Add test for concat_text method

examples/text.rs

@@ -18,10 +18,10 @@ fn main() {
                 }
             }
         }
-        for line in flow.lines {
-            for w in line.words {
-                println!("{}", w.text);
-            }
-        }
+        // for line in flow.lines {
+        //     for w in line.words {
+        //         println!("{}", w.text);
+        //     }
+        // }
     // }
 }

src/text.rs

@@ -7,12 +7,15 @@ use crate::{util::avg, flow::{Word, Rect}};
 
 pub fn concat_text<'a, E: Encoder + 'a>(out: &mut String, items: impl Iterator<Item=&'a TextSpan<E>> + Clone) -> Vec<Word> {
     let mut words: Vec<Word> = vec![];
-    // dbg!(items.clone().map(|s| s).collect::<Vec<_>>());
-    // gaps between each char
+
+    // Calculate gaps between each char, the unit is em, relative to the font size.
     let gaps = items.clone()
         .flat_map(|s| {
+            // the transform matrix is from em space to device space
+            // so we need to invert it
             let tr_inv = s.transform.matrix.inverse();
             let pos = (tr_inv * s.transform.vector).x();
+
             s.chars.iter()
                 .filter(|c| !s.text[c.offset..].chars().next().unwrap().is_whitespace())
                 .map(move |c| (c.pos + pos, c.pos + pos + c.width, s.font_size))
@@ -26,6 +29,7 @@ pub fn concat_text<'a, E: Encoder + 'a>(out: &mut String, items: impl Iterator<I
     let space_gap = (0.5 * font_size).min(2.0 * avg(gaps).unwrap_or(0.0)); //2.0 * gaps[gaps.len()/2];
 
     let mut end = 0.; // trailing edge of the last char
+    // out中最后一个字符是否是空格
     let mut trailing_space = out.chars().last().map(|c| c.is_whitespace()).unwrap_or(true);
     let mut word_start_pos = 0.0;
     let mut word_start_idx = out.len();
@@ -38,16 +42,19 @@ pub fn concat_text<'a, E: Encoder + 'a>(out: &mut String, items: impl Iterator<I
         let mut pos = 0; // byte index of last char into span.text
         let tr_inv = span.transform.matrix.inverse();
         let x_off = (tr_inv * span.transform.vector).x();
+
         for c in span.chars.iter() {
-
+            // current string of TextChar
             let s = &span.text[pos..c.offset];
             if c.offset > 0 {
                 let is_whitespace = s.chars().all(|c| c.is_whitespace());
+                // 在不为空格的时候， 将 s 写入 out.
                 if !trailing_space || !is_whitespace {
                     out.extend(s.nfkc());
                 }
                 trailing_space = is_whitespace;
             }
+            // 在 s 不为空格，且有gap 的时候，记录一个 word.
             if !trailing_space && c.pos + x_off > end + space_gap {
                 words.push(Word {
                     text: out[word_start_idx..].into(),
@@ -80,6 +87,7 @@ pub fn concat_text<'a, E: Encoder + 'a>(out: &mut String, items: impl Iterator<I
                 y_max = y_max.max(span.rect.max_y());
             }
         }
+
         trailing_space = span.text[pos..].chars().all(|c| c.is_whitespace());
 
         out.extend(span.text[pos..].nfkc());
@@ -95,4 +103,51 @@ pub fn concat_text<'a, E: Encoder + 'a>(out: &mut String, items: impl Iterator<I
     });
 
     words
+}
+
+#[cfg(test)]
+mod tests {
+    use pathfinder_geometry::{rect::RectF, transform2d::Transform2F};
+    use pdf_render::{font::OutlineBuilder, Fill, TextChar};
+
+    use super::*;
+
+    #[test]
+    fn test_concat_text() {
+        let text_span: TextSpan<OutlineBuilder> = TextSpan {
+            rect: RectF::from_points(Vector2F::new(56.8, 55.85077), Vector2F::new(136.26399, 67.85077)),
+            width: 79.464,
+            bbox: None,
+            font_size: 12.0,
+            font: None,
+            text: "hello world".to_string(),
+            chars: vec![
+                TextChar { offset: 0, pos: 0.0, width: 7.224001 },
+                TextChar { offset: 1, pos: 7.224001, width: 7.224001 },
+                TextChar { offset: 2, pos: 14.448002, width: 7.224001 },
+                TextChar { offset: 3, pos: 21.672003, width: 7.224001 },
+                TextChar { offset: 4, pos: 28.896004, width: 7.224001 },
+                TextChar { offset: 5, pos: 36.120003, width: 7.224001 },
+                TextChar { offset: 6, pos: 43.344, width: 7.224001 },
+                TextChar { offset: 7, pos: 50.568, width: 7.224001 },
+                TextChar { offset: 8, pos: 57.792, width: 7.224001 },
+                TextChar { offset: 9, pos: 65.016, width: 7.224001 },
+                TextChar { offset: 10, pos: 72.24, width: 7.224001 },
+            ],
+            color: Fill::Solid(0.0, 0.5019608, 0.0),
+            alpha: 1.0,
+            transform: Transform2F::row_major(1.0, 0.0, 56.8, 0.0, 1.0, 67.85077),
+            mode: pdf::content::TextMode::Fill,
+            op_nr: 18,
+        };
+
+        let mut output = String::new();
+        let words = concat_text(&mut output, vec![&text_span].into_iter());
+
+        // Assert the concatenated text
+        assert_eq!(output, "hello world");
+
+        // Assert the words
+        assert_eq!(words.len(), 2); // Expect two words: "hello" and "world"
+    }
 }